Switch in equations using ContinuousCallback is handled by Rodas5() but not QNDF()

My apologies. This shows that x_1 approaches the expected value 49.50 as I reduce the tolerance. The differential state still remains outside of tolerance, however.

1e-8:

u = [49.479690908110484, 13.682966651643868, 0.5, 49.99999999999998]

1e-9:

u = [49.4915514356334, 13.685537303140494, 0.5, 49.99999999999999]

1e-10:

u = [49.494069541722695, 13.686082918800729, 0.5, 49.99999999999998]

1e-11:

u = [49.49822477214678, 13.686983072701429, 0.5, 49.999999999999986]

1e-12:

u = [49.49990766890697, 13.687347610721506, 0.5, 49.99999999999997]

I saw Yingbo’s BDF method derivation reference in his blog. Do you recommend any references for understanding of algebraic state interpolation?